Using Large Language Models locally

Using Large Language Models locally

05 December 2024

FWO - HOEK 38

Artificial Inteligence
HPC
ELIXIR
live training

Using Large Language Models locally

Target Audience:
VIB PhD student
VIB postdoc
VIB staff scientist
VIB group leader or expert
VIB administrative support
VIB technical support
Flemish academic or researcher
Non-Flemish academic or researcher
Industry professional
Location:

FWO - HOEK 38

Duration:

one afternoon

General context

Although most of us use Large Language Models via a web interface, this is not always the ideal scenario. GDPR or intellectual property considerations may prohibit exposing data or even queries to providers of such services. A model may have to be fine-tuned to increase the quality for certain applications. In such circumstances, you may want to deploy an LLM locally, and this training will show you how to do that and give some illustrations of applications you can implement.

Learning outcomes

When you follow this training you will be able to:

  • install and run LLMs on local hardware;
  • use Retrieval Augmented Generation (RAG) to chat with your own documents/data;
  • fine-tune an LLM to better answer your questions;
  • optimize an LLM to reduce requirements for inference.
Required skills

Familiarity with Python will be required, experience using the Bash command line will help.

Trainers

Program

0900-17:30 - Local LLMs

Practical info

Location & Venue

05 December 2024

Brussels - HOEK 38

Leuvenseweg 38
1000 Brussel
Belgium

Public transport

05 December 2024

Brussels - HOEK 38
Public transport

No public transport information has been provided for this location.

Bike

05 December 2024

Brussels - HOEK 38
Route description

05 December 2024

Brussels - HOEK 38
Parking

No parking information has been provided for this location.

Venue contact

05 December 2024

Brussels - HOEK 38
Location contact

(+32) 251 291 10
hoek38@fwo.be